learning_rate: 5e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 10
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer
5.5901	0.6186	300	1.1050	0.2548
1.7943	1.2371	600	0.9902	0.2170
1.4770	1.8557	900	0.9139	0.2426
1.1331	2.4742	1200	0.8916	0.1998
0.9384	3.0928	1500	0.9659	0.2013
0.7179	3.7113	1800	0.9658	0.1939
0.5310	4.3299	2100	1.0882	0.1956
0.4004	4.9485	2400	1.1178	0.1984
0.2511	5.5670	2700	1.3112	0.1985
0.1849	6.1856	3000	1.3949	0.1990
0.1180	6.8041	3300	1.5543	0.1959
0.0772	7.4227	3600	1.6783	0.1980
0.0586	8.0412	3900	1.8147	0.1977
0.0330	8.6598	4200	1.9958	0.1934
0.0213	9.2784	4500	2.0510	0.1942
0.0137	9.8969	4800	2.0989	0.1934

dhasmana/Rajasthani-Magadhi-Chhattisgarhi-Bhojpuri-w2v-bert-2.0

Rajasthani-Magadhi-Chhattisgarhi-Bhojpuri-w2v-bert-2.0